19
Using Request Queues for Enhancing the Performance of Operations in Smart Homes Andreas Kamilaris and Andreas Pitsillides Networks Research Laboratory Dept. of Computer Science, University of Cyprus 7 th Performance Monitoring, Measurement and Evaluation of Heterogeneous Wireless and Wired Networks (PM2HW2N) Workshop Paphos, Cyprus 21 st October 2012

Using Request Queues for Enhancing the Performance of Operations in Smart Homes

Embed Size (px)

DESCRIPTION

This is the presentation for the research paper "Using Request Queues for Enhancing the Performance of Operations in Smart Buildings". It has been presented during the 7th ACM International Workshop on Performance Monitoring, Measurement and Evaluation of Heterogeneous Wireless and Wired Networks (PM2HW2N), in Proc. of MSWiM 2012, in Paphos, Cyprus, October 2012. Its abstract is as follows: Modern houses and buildings are being equipped with embedded wireless sensors and actuators, offering advanced automation possibilities. Embedded technology is becoming mature, forming an enticing option for real-life deployments. Still, embedded wireless computing does not constitute a guaranteed reliable solution since transmission failures occur in the wireless medium while resource-constrained devices have battery limitations and frequent failures. In this paper, we examine the use of request queues as a mechanism to manage the communication with embedded devices. Located in a middleware application framework for smart homes, request queues offer enhanced reliability and fault tolerance, supporting multiple tenants simultaneously.

Citation preview

Page 1: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

Using Request Queues for Enhancing the Performance of Operations in Smart Homes

Andreas Kamilaris and Andreas Pitsillides

Networks Research Laboratory Dept. of Computer Science, University of Cyprus

7th Performance Monitoring, Measurement and Evaluation of Heterogeneous Wireless and

Wired Networks (PM2HW2N) Workshop

Paphos, Cyprus 21st October 2012

Page 2: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

University of CyprusIntroduction

• Modern smart homes tend to being equipped with embedded sensors, actuators, smart power outlets and meters.

• Transmission failures are a common happening in indoor environments.

• Resource-constrained home devices have battery limitations and frequent failures.

• Reduced reliability and performance.• No guarantees.• High unpredictability.

Page 3: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

University of CyprusMotivation

• Better management of the interactions with embedded devices.

• Reliability and performance need to be ensured.• Intermediate entities for handling communication with the

smart home environment.• Request queues a suitable intermediate data structure for

enhancing the performance of pervasive applications that target smart homes and building automation.

Page 4: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

University of CyprusRequest Queues

• Request queues are defined as FIFO queues.• Installed on middleware applications for smart buildings.• Handle the requests coming from the building's tenants,

targeting the embedded devices of the building.

Page 5: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

University of CyprusAn Application Framework for Smart Buildings

• Each home device is associated with its own request queue.

Page 6: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

University of CyprusRequest Queue Operation

Page 7: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

University of CyprusRequest Queue Analysis

• Incoming requests need to wait in the queue for their turn, in order to be executed.

• The arrival rate lambda of Web clients at the framework is modeled by the exponential distribution.

Page 8: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

University of CyprusExperimental Setup• 6LoWPAN-enabled Telosb sensor motes.• Sensing capabilities exposed as RESTful Web services.• Transmission power at -25 dBm, message sizes 128 bytes.

Page 9: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

University of CyprusRequest Queue Analysis

• Request queue retransmission interval α

Page 10: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

University of CyprusRequest Queue Analysis

• Influence of transmission failures and different arrival rates on retransmission interval α.

Page 11: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

University of CyprusRequest Queue Analysis

• Influence of varied response times on retransmission interval α.

• RTT times and St. Deviation values learned from the device thread. • Set initially to a larger value, leaving a "safe margin”.• Fine-tune adaptively during the device operation.

Page 12: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

University of CyprusPotential Benefits

• Multi-Client Support

• Multi-hop topology additional delays of around 200 ms.• Heavy workload increases response times by 18-20% in the single-

hop case and 14-17% in the multi-hop topology.

Page 13: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

University of CyprusPotential Benefits

• Avoiding Transmission Failures

• In light workload, transmission failures not affect significantly the response times.

• In heavy workloads, transmission failures cause the response times to grow almost exponentially.

Page 14: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

University of CyprusPotential Benefits

• Estimating Potential Response Times

• Average estimation error is 12.38%, and it increases to 14.60% when the request queue size becomes larger than 2.

Page 15: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

University of CyprusPotential Benefits

• Load Balancing for Serving High Traffic

• In low traffic, the improvement of performance is around 4-6%.• In increased traffic, the improvement reaches 11%.

Page 16: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

University of CyprusPotential Benefits

• Handling Priorities

1. Assign a priority to each request (low, normal or high).

2: Each prioritized request is translated into an integer value by the application framework, according to the following formula: low=1, normal=5 and high=10.

3: The priority heap selects for execution the request with the highest priority number.

4: To avoid starvation of low-priority requests, the priorities of all waiting requests are increased by 1 at every round, i.e. at a successful execution of some request.

Page 17: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

University of CyprusFuture Work

• Support for moving objects and dynamic environments

• Detailed system analysis by employing queuing theory.

• Task scheduling through priority handling.

• Device-centric Vs Service-centric approach.

• Device Utilization.

• Expected number of building tenants/workers and waiting times.

• Estimation of the probability of the request queue to be in certain state.

• Distribution of the length of the busy periods, sojourn and waiting times.

• Distribution of the number of arrivals during the service time etc.

Page 18: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

University of CyprusConclusions

• Intermediate request queues a promising mechanism.• Enhancing pervasive applications with numerous benefits.• More useful in scenarios with multiple, concurrent users and

increased percentage of transmission failures.• Support of prioritized requests.• Serve high traffic through load balancing.

• A dynamic, adaptive system that handles failures in the embedded environment providing certain reliability.

Page 19: Using Request Queues for Enhancing the Performance of Operations in Smart Homes

Thanks for your attention!

Contact Details: Andreas Kamilaris ([email protected])